Commit eb536fa
committed
fix(tests): seed reorder folders outside Q_ASSERT (stripped in Release CI)
The real cause of the CI-only test_two_columns_node_explorer_sort failure
(NOT a reorder race). bringUpExplorer wrapped the seed createFolder/createFile
calls inside Q_ASSERT(!...isEmpty()). CI builds Release (CMAKE_BUILD_TYPE
defaults to Release in the root CMakeLists; ci-win.yml passes
-DCMAKE_BUILD_TYPE=Release), where QT_NO_DEBUG strips Q_ASSERT together with
its argument expression. The seeding was therefore compiled out on
Linux/Windows CI, leaving the notebook empty; the subsequent reorder saw
SetChildrenOrder current=0 vs submitted=2 -> VXCORE_ERR_PERMUTATION_MISMATCH
-> reorderCompleted(false) -> errorOccurred (never nodesReordered) ->
reorderedSpy.wait(5000) timed out. Green in local Debug and on macOS (built
with assertions), red on Release CI.
Confirmed from instrumented CI logs: zero createFolder calls before the
reorder, cache_size=1 with an empty root config. Move the four seed calls to
unconditional statements; keep a Q_ASSERT sanity check on the stored ids
(Q_UNUSED to avoid Release unused-variable warnings).
Also restores the vxcore submodule pointer to a clean (non-instrumented)
commit after the diagnostic build.1 parent 0f4d553 commit eb536fa
2 files changed
Lines changed: 19 additions & 5 deletions
| Original file line number | Diff line number | Diff line change | |
|---|---|---|---|
| |||
155 | 155 | | |
156 | 156 | | |
157 | 157 | | |
158 | | - | |
159 | | - | |
160 | | - | |
161 | | - | |
| 158 | + | |
| 159 | + | |
| 160 | + | |
| 161 | + | |
| 162 | + | |
| 163 | + | |
| 164 | + | |
| 165 | + | |
| 166 | + | |
| 167 | + | |
| 168 | + | |
| 169 | + | |
| 170 | + | |
| 171 | + | |
| 172 | + | |
| 173 | + | |
| 174 | + | |
| 175 | + | |
162 | 176 | | |
163 | 177 | | |
164 | 178 | | |
| |||
0 commit comments